SNAP (MSSA) — journal-club digest

Personal study digest for a new ID fellow. Findings are from the cited trial; the appraisal, Stats Corner, and wider-literature notes are reviewer synthesis. Not a substitute for the paper.

Bottom line

Cefazolin was noninferior to flucloxacillin/cloxacillin for 90-day mortality in MSSA bacteremia — and caused less acute kidney injury.

The question: which IV β-lactam backbone is better for penicillin-resistant MSSA bacteremia?

PICO

Population Adults ≥18 with penicillin-resistant MSSA bacteremia, enrolled ≤72 h of first positive culture. Excluded: maintenance dialysis, polymicrobial bacteremia, penicillin/cefazolin allergy.

Intervention Cefazolin 2 g IV q8h (q6h if critically ill or endocarditis), renally adjusted.

Comparison Antistaphylococcal penicillin — flucloxacillin 2 g IV q6h or cloxacillin 2 g IV q4h.

Outcome Primary: all-cause death by 90 days. Key secondary (safety): acute kidney injury ≤14 days.

Why this trial mattered

S. aureus bacteremia kills more than 1 in 4 patients within 90 days, yet the choice of β-lactam backbone rested on expert opinion. Endocarditis guidelines favored antistaphylococcal penicillins over cefazolin because of the theoretical cefazolin inoculum effect — type A β-lactamase can hydrolyze cefazolin at high organism burden in vitro, raising fear of failure in endocarditis and other high-inoculum disease. Observational data pulled the other way, favoring cefazolin on mortality and tolerability. SNAP is the first randomized head-to-head, so it settles a decades-old, purely opinion-based split.

Design at a glance

Design RCT, open-label Framework Bayesian adaptive platform Randomized 1,341 (1,287 analyzed) Allocation 1:1, no response-adaptive Primary 90-day all-cause death NI margin adjusted OR < 1.2 Ran Feb 2022 – Aug 2024

What they found

90-day mortality: cefazolin 15.0% (97/645) vs. antistaphylococcal penicillin 17.0% (109/642) — adjusted OR 0.81 (95% CrI 0.59–1.12); probability of noninferiority 99.2%, of superiority 89.8%. Noninferiority met; superiority not declared.
Acute kidney injury (≤14 d): 13.9% (92/660) vs. 19.6% (127/648) — adjusted OR 0.67 (95% CrI 0.50–0.89); probability of superiority 99.7%; probability of lower renal-replacement initiation 94.6%.
Endocarditis subgroup: consistent (adjusted OR 0.54; 95% CrI 0.20–1.43) — reassuring against the inoculum-effect concern.
Tolerability: fewer serious adverse reactions (adjusted OR 0.41; 0.21–0.77) and far fewer discontinuations for adverse events (adjusted OR 0.21; 0.11–0.38) with cefazolin.
Conduct: stopped when noninferiority was met; 4.0% lost or withdrawn.

Bottom line & practice change

For most adults with penicillin-resistant MSSA bacteremia, cefazolin is the preferred backbone: equal 90-day survival, less AKI, fewer serious adverse reactions, fewer discontinuations, and convenient q8h dosing. Reserve an antistaphylococcal penicillin for when cefazolin is contraindicated or inoculum-effect concern is high (e.g., some CNS infections not tested here). Two caveats travel with that: the non-mortality outcomes are open-label, and US practice uses nafcillin/oxacillin, which requires class extrapolation.

How good is this evidence?

Strengths

First RCT to answer the question; ends expert-opinion practice.
Prespecified NI margin and published Bayesian stopping rules.
Objective primary endpoint (all-cause death) resists open-label bias.
Pragmatic and generalizable; ID consult >99%; large (n≈1,300).
Consistent across sensitivity analyses and the endocarditis subgroup.

Weaknesses

Open-label — no blinding of assignment.
AKI captured only through day 14.
No fixed N; stopping at a boundary can flatter point estimates.
Flucloxacillin/cloxacillin tested, not US nafcillin/oxacillin.
Limited analytic reproducibility (no public code/dataset).

Bias / threats

Open-label can bias soft outcomes (AKI, discontinuations), not death.
Dialysis patients excluded from the renal comparison.
Competing risks: sicker comparator patients may shape AKI counts.
Early-stop optimism for the superiority-flavored signals.

Note the asymmetry: the randomized, objective mortality result is robust; the softer secondary signals (AKI, discontinuations) deserve more caution.

Stats Corner: Bayesian noninferiority in an adaptive platform

SNAP doesn’t report a p-value. It reports a posterior probability — given the data and the model, the chance the true effect beats a threshold. Noninferiority was declared when the probability that the adjusted OR is below the prespecified margin of 1.2 exceeded 0.99 (here it reached 99.2%). The interval quoted is a 95% credible interval — the range the OR sits in with 95% probability, a more direct statement than a frequentist confidence interval. Because the mortality CrI (0.59–1.12) still crosses 1.0, superiority is not declared even though noninferiority is comfortably met — the two questions are separate. The platform design runs many comparisons at once (patients enter a silo by methicillin status, then are randomized within domains); allocation here was fixed 1:1, not response-adaptive. One caution: stopping when a boundary is crossed can make the point estimates (OR 0.81; AKI 0.67) look a touch more favorable than a fixed-N trial would.

Where it sits in the literature

Reviewer synthesis (no editorial accompanied this copy) — kept separate from the trial’s own results.

Observational meta-analysis had already favored cefazolin for 30-day mortality (OR 0.73, 95% CI 0.62–0.85, cited in the trial’s introduction); SNAP now confirms that direction with randomization, removing the confounding-by-indication that made ASPs look worse (sicker and endocarditis patients preferentially got them). It challenges the longstanding endocarditis-guideline preference for antistaphylococcal penicillins rooted in the inoculum-effect concern. Keep two adjacent questions separate for the discussion: early IV-to-oral step-down in low-risk SAB (SABATO, Lancet Infect Dis 2024) and the other SNAP domains (e.g., adjunctive therapy) — SNAP’s platform protocol is published (Clin Infect Dis 2022). This backbone result speaks only to which IV β-lactam to choose.

Sources: SNAP Trial Group. Cefazolin for MSSA Bacteremia. N Engl J Med 2026;394:2329-2339 (doi:10.1056/NEJMoa2506905; NCT05137119) — article, supplementary appendix, and protocol on file. Interim-analysis timeline from the trial’s public DSMC letters and Statistical Implementation Guides. Observational 30-day mortality estimate (OR 0.73) cited in the trial’s introduction. Related: SABATO (Lancet Infect Dis 2024); SNAP platform protocol (Clin Infect Dis 2022).